Paperless isn’t a very complicated app, but there are a few components, so some basic documentation is in order. If you follow along in this document and still have trouble, please open an issue on GitHub so I can fill in the gaps.


The source is currently only available via GitHub, so grab it from there, either by using git:

$ git clone
$ cd paperless

or just download the tarball and go that route:

$ cd to the directory where you want to run Paperless
$ wget
$ unzip
$ cd paperless-master

Installation & Configuration

You can go multiple routes with setting up and running Paperless. The Vagrant route is quick & easy, but means you’re running a VM which comes with memory consumption etc. We also support Docker, which you can use natively under Linux and in a VM with Docker Machine (this guide was written for native Docker usage under Linux, you might have to adapt it for Docker Machine.) Not to forget the virtualenv, this is similar to bare metal with the exception that you have to activate the virtualenv first. Last but not least, the standard bare metal approach is a little more complicated, but worth it because it makes it easier should you want to contribute some code back.

Standard (Bare Metal)

  1. Install the requirements as per the requirements page.

  2. Within the extract of go to the src directory.

  3. Copy paperless.conf.example to /etc/paperless.conf also the virtual envrionment look there for it and open it in your favourite editor. Because this file contains passwords it should only be readable by user root and paperless ! Set the values for:

    • PAPERLESS_CONSUMPTION_DIR: this is where your documents will be dumped to be consumed by Paperless.
    • PAPERLESS_PASSPHRASE: this is the passphrase Paperless uses to encrypt/decrypt the original document.
    • PAPERLESS_OCR_THREADS: this is the number of threads the OCR process will spawn to process document pages in parallel.
  4. Initialise the SQLite database with ./ migrate.

  5. Create a user for your Paperless instance with ./ createsuperuser. Follow the prompts to create your user.

  6. Start the webserver with ./ runserver <IP>:<PORT>. If no specifc IP or port are given, the default is also known as http://localhost:8000/. You should now be able to visit your (empty) at Paperless webserver or whatever you chose before. You can login with the user/pass you created in #5.

  7. In a separate window, change to the src directory in this repo again, but this time, you should start the consumer script with ./ document_consumer.

  8. Scan something or put a file into the CONSUMPTION_DIR.

  9. Wait a few minutes

  10. Visit the document list on your webserver, and it should be there, indexed and downloadable.

Vagrant Method

  1. Install Vagrant. How you do that is really between you and your OS.

  2. Run vagrant up. An instance will start up for you. When it’s ready and provisioned...

  3. Run vagrant ssh and once inside your new vagrant box, edit /etc/paperless.conf and set the values for:

    • PAPERLESS_CONSUMPTION_DIR: this is where your documents will be dumped to be consumed by Paperless.
    • PAPERLESS_PASSPHRASE: this is the passphrase Paperless uses to encrypt/decrypt the original document.
    • PAPERLESS_SHARED_SECRET: this is the “magic word” used when consuming documents from mail or via the API. If you don’t use either, leaving it blank is just fine.
  4. Exit the vagrant box and re-enter it with vagrant ssh again. This updates the environment to make use of the changes you made to the config file.

  5. Initialise the database with /opt/paperless/src/ migrate.

  6. Still inside your vagrant box, create a user for your Paperless instance with /opt/paperless/src/ createsuperuser. Follow the prompts to create your user.

  7. Start the webserver with /opt/paperless/src/ runserver You should now be able to visit your (empty) Paperless webserver at You can login with the user/pass you created in #6.

  8. In a separate window, run vagrant ssh again, but this time once inside your vagrant instance, you should start the consumer script with /opt/paperless/src/ document_consumer.

  9. Scan something. Put it in the CONSUMPTION_DIR.

  10. Wait a few minutes

  11. Visit the document list on your webserver, and it should be there, indexed and downloadable.

Docker Method

  1. Install Docker.


    As mentioned earlier, this guide assumes that you use Docker natively under Linux. If you are using Docker Machine under Mac OS X or Windows, you will have to adapt IP addresses, volume-mounting, command execution and maybe more.

  2. Install docker-compose. [1]


    If you want to use the included docker-compose.yml.example file, you need to have at least Docker version 1.10.0 and docker-compose version 1.6.0.

    See the Docker installation guide on how to install the current version of Docker for your operating system or Linux distribution of choice. To get an up-to-date version of docker-compose, follow the docker-compose installation guide if your package repository doesn’t include it.

  3. Create a copy of docker-compose.yml.example as docker-compose.yml and a copy of docker-compose.env.example as docker-compose.env. You’ll be editing both these files: taking a copy ensures that you can git pull to receive updates without risking merge conflicts with your modified versions of the configuration files.

  4. Modify docker-compose.yml to your preferences, following the instructions in comments in the file. The only change that is a hard requirement is to specify where the consumption directory should mount.

  5. Modify docker-compose.env and adapt the following environment variables:


    This is the passphrase Paperless uses to encrypt/decrypt the original document.


    This is the number of threads the OCR process will spawn to process document pages in parallel. If the variable is not set, Python determines the core-count of your CPU and uses that value.


    If you want the OCR to recognize other languages in addition to the default English, set this parameter to a space separated list of three-letter language-codes after ISO 639-2/T. For a list of available languages – including their three letter codes – see the Debian packagelist.


    If you want to mount the consumption volume (directory /consume within the containers) to a host-directory – which you probably want to do – access rights might be an issue. The default user and group paperless in the containers have an id of 1000. The containers will enforce that the owning group of the consumption directory will be paperless to be able to delete consumed documents. If your host-system has a group with an ID of 1000 and you don’t want this group to have access rights to the consumption directory, you can use USERMAP_GID to change the id in the container and thus the one of the consumption directory. Furthermore, you can change the id of the default user as well using USERMAP_UID.

  6. Run docker-compose up -d. This will create and start the necessary containers.

  7. To be able to login, you will need a super user. To create it, execute the following command:

    $ docker-compose run --rm webserver createsuperuser

    This will prompt you to set a username (default paperless), an optional e-mail address and finally a password.

  8. The default docker-compose.yml exports the webserver on your local port 8000. If you haven’t adapted this, you should now be able to visit your Paperless webserver at You can login with the user and password you just created.

  9. Add files to consumption directory the way you prefer to. Following are two possible options:

    1. Mount the consumption directory to a local host path by modifying your docker-compose.yml:

      diff --git a/docker-compose.yml b/docker-compose.yml
      --- a/docker-compose.yml
      +++ b/docker-compose.yml
      @@ -17,9 +18,8 @@ services:
                   - paperless-data:/usr/src/paperless/data
                   - paperless-media:/usr/src/paperless/media
      -            - /consume
      +            - /local/path/you/choose:/consume


      While the consumption container will ensure at startup that it can delete a consumed file from a host-mounted directory, it might not be able to read the document in the first place if the access rights to the file are incorrect.

      Make sure that the documents you put into the consumption directory will either be readable by everyone (chmod o+r file.pdf) or readable by the default user or group id 1000 (or the one you have set with USERMAP_UID or USERMAP_GID respectively).

    2. Use docker cp to copy your files directly into the container:

      $ # Identify your containers
      $ docker-compose ps
              Name                       Command                State     Ports
      paperless_consumer_1    /sbin/ ...   Exit 0
      paperless_webserver_1   /sbin/ ...   Exit 0
      $ docker cp /path/to/your/file.pdf paperless_consumer_1:/consume

      docker cp is a one-shot-command, just like cp. This means that every time you want to consume a new document, you will have to execute docker cp again. You can of course automate this process, but option 1 is generally the preferred one.


      docker cp will change the owning user and group of a copied file to the acting user at the destination, which will be root.

      You therefore need to ensure that the documents you want to copy into the container are readable by everyone (chmod o+r file.pdf) before copying them.

[1]You of course don’t have to use docker-compose, but it simplifies deployment immensely. If you know your way around Docker, feel free to tinker around without using compose!

Making Things a Little more Permanent

Once you’ve tested things and are happy with the work flow, you can automate the process of starting the webserver and consumer automatically.

Standard (Bare Metal, Systemd)

If you’re running on a bare metal system that’s using Systemd, you can use the service unit files in the scripts directory to set this up. You’ll need to create a user called paperless (without login (if not already done so #5)) and setup Paperless to be in a place that this new user can read and write to. Be sure to edit the service scripts to point to the proper location of your paperless install, referencing the appropriate Python binary. For example: ExecStart=/path/to/python3 /path/to/paperless/src/ document_consumer. If you don’t want to make a new user, you can change the Group and User variables accordingly.

Then, as root (or using sudo) you can just copy the .service files to the Systemd directory and tell it to enable the two services:

# cp /path/to/paperless/scripts/paperless-consumer.service /etc/systemd/system/
# cp /path/to/paperless/scripts/paperless-webserver.service /etc/systemd/system/
# systemctl enable paperless-consumer
# systemctl enable paperless-webserver
# systemctl start paperless-consumer
# systemctl start paperless-webserver

Ubuntu 14.04 (Bare Metal, Upstart)

Ubuntu 14.04 and earlier use the Upstart init system to start services during the boot process. To configure Upstart to run Paperless automatically after restarting your system:

  1. Change to the directory where Upstart’s configuration files are kept: cd /etc/init

  2. Create a new file: sudo nano paperless-server.conf

  3. In the newly-created file enter:

    start on (local-filesystems and net-device-up IFACE=eth0)
    stop on shutdown
    respawn limit 10 5
      exec /srv/paperless/src/ runserver
    end script

    Note that you’ll need to replace /srv/paperless/src/ with the path to the script in your installation directory.

If you are using a network interface other than eth0, you will have to change IFACE=eth0. For example, if you are connected via WiFi, you will likely need to replace eth0 above with wlan0. To see all interfaces, run ifconfig -a.

Save the file.

  1. Create a new file: sudo nano paperless-consumer.conf

  2. In the newly-created file enter:

    start on (local-filesystems and net-device-up IFACE=eth0)
    stop on shutdown
    respawn limit 10 5
      exec /srv/paperless/src/ document_consumer
    end script
Replace /srv/paperless/src/ with the same values as in step 3 above and replace eth0 with the appropriate value, if necessary. Save the file.

These two configuration files together will start both the Paperless webserver and document consumer processes when the file system and network interface specified is available after boot. Furthermore, if either process ever exits unexpectedly, Upstart will try to restart it a maximum of 10 times within a 5 second period.

Using a Real Webserver

The default is to use Django’s development server, as that’s easy and does the job well enough on a home network. However, if you want to do things right, it’s probably a good idea to use a webserver capable of handling more than one thread. You will also have to let the webserver serve the static files (CSS, JavaScript) from the directory configured in PAPERLESS_STATICDIR. For that, you need to run ./ collectstatic in the src directory. The default static files directory is ../static.


This is a configuration supplied by steckerhalter on GitHub. It uses Apache and mod_wsgi, with a Paperless installation in /home/paperless/:

<VirtualHost *:80>

    Alias /static/ /home/paperless/paperless/static/
    <Directory /home/paperless/paperless/static>
        Require all granted

    WSGIScriptAlias / /home/paperless/paperless/src/paperless/
    WSGIDaemonProcess user=paperless group=paperless threads=5 python-path=/home/paperless/paperless/src:/home/paperless/.env/lib/python3.4/site-packages

    <Directory /home/paperless/paperless/src/paperless>
            Require all granted

Nginx + Gunicorn

If you’re using Nginx, the most common setup is to combine it with a Python-based server like Gunicorn so that Nginx is acting as a proxy. Below is a copy of a simple Nginx configuration fragment making use of SSL and IPv6 to refer to a gunicorn instance listening on a local Unix socket:

upstream transfer_server {
  server unix:/run/ fail_timeout=0;

# Redirect requests on port 80 to 443
server {
  listen 80;
  listen [::]:80;
  rewrite ^ https://$server_name$request_uri? permanent;

server {

  listen 443 ssl;
  listen [::]:443;
  client_max_body_size 4G;
  keepalive_timeout 5;
  root /var/www/;

  ssl on;

  ssl_certificate         /etc/letsencrypt/live/;
  ssl_certificate_key     /etc/letsencrypt/live/;
  ssl_trusted_certificate /etc/letsencrypt/live/;
  ssl_session_timeout 1d;
  ssl_session_cache shared:SSL:50m;

  # Diffie-Hellman parameter for DHE ciphersuites, recommended 2048 bits
  # Generate with:
  #   openssl dhparam -out /etc/nginx/dhparam.pem 2048
  ssl_dhparam /etc/nginx/dhparam.pem;

  # What Mozilla calls "Intermediate configuration"
  # Copied from
  ssl_protocols TLSv1 TLSv1.1 TLSv1.2;
  ssl_prefer_server_ciphers on;

  add_header Strict-Transport-Security max-age=15768000;

  ssl_stapling on;
  ssl_stapling_verify on;

  access_log /var/log/nginx/ main;
  error_log /var/log/nginx/ info;

  location / {
    try_files $uri @proxy_to_app;

  location @proxy_to_app {
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    proxy_set_header X-Forwarded-Proto https;
    proxy_set_header Host $host;
    proxy_redirect off;
    proxy_pass http://transfer_server;


Once you’ve got Nginx configured, you’ll want to have a configuration file for your gunicorn instance. This should do the trick:

import os

bind = 'unix:/run/'
backlog = 2048
workers = 6
worker_class = 'sync'
worker_connections = 1000
timeout = 30
keepalive = 2
debug = False
spew = False
daemon = False
pidfile = None
umask = 0
user = None
group = None
tmp_upload_dir = None
errorlog = '/var/log/'
loglevel = 'warning'
accesslog = '/var/log/'
proc_name = None

def post_fork(server, worker):"Worker spawned (pid: %s)",

def pre_fork(server, worker):

def pre_exec(server):"Forked child, re-executing.")

def when_ready(server):"Server is ready. Spawning workers")

def worker_int(worker):"worker received INT or QUIT signal")

    ## get traceback info
    import threading, sys, traceback
    id2name = dict([(th.ident, for th in threading.enumerate()])
    code = []
    for threadId, stack in sys._current_frames().items():
        code.append("\n# Thread: %s(%d)" % (id2name.get(threadId,""),
        for filename, lineno, name, line in traceback.extract_stack(stack):
            code.append('File: "%s", line %d, in %s' % (filename,
                lineno, name))
            if line:
                code.append("  %s" % (line.strip()))

def worker_abort(worker):"worker received SIGABRT signal")


You may use the Ubuntu explanation above. Replace (local-filesystems and net-device-up IFACE=eth0) with vagrant-mounted.


If you’re using Docker, you can set a restart-policy in the docker-compose.yml to have the containers automatically start with the Docker daemon.